Index usage for retrieving data without filter predicate

Question

Hi,
does someone have a explanation for the following scenario:
I have a table T1 with an index OID_IX on column ( object_id ) - The table is a CTAS from dba_objects just to populate it with data.
There are no other Indexes present. The table and the index are analyzed !
When I run the following query the table is accessed FULL ( not using the index )
SELECT OBJECT_ID FROM T1;
SQL> select object_id from t1;
485984 rows selected.
Elapsed: 00:00:01.76
Execution Plan
Plan hash value: 3617692013
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 485K| 2372K| 1528 (1)| 00:00:19 |
| 1 | TABLE ACCESS FULL| T1 | 485K| 2372K| 1528 (1)| 00:00:19 |
Statistics
1 recursive calls
0 db block gets
7396 consistent gets
0 physical reads
0 redo size
2887158 bytes sent via SQL*Net to client
5684 bytes received via SQL*Net from client
487 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
485984 rows processed
But if I add a predicate ( even though it is useless in this case ) the index is taken and the query runs faster:
JDBC@toekb> select object_id from t1 where object_id != -999;
485960 rows selected.
Elapsed: 00:00:01.40
Execution Plan
Plan hash value: 3555700789
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 485K| 2372K| 242 (3)| 00:00:03 |
|* 1 | INDEX FAST FULL SCAN| OID_IX | 485K| 2372K| 242 (3)| 00:00:03 |
Predicate Information (identified by operation id):
1 - filter("OBJECT_ID"<>(-999))
Statistics
1 recursive calls
0 db block gets
1571 consistent gets
0 physical reads
0 redo size
2766124 bytes sent via SQL*Net to client
5684 bytes received via SQL*Net from client
487 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
485960 rows processed
here my setup:
sqlsql--
drop table t1 purge ;
create table t1 tablespace users as select * from dba_objects;
insert into t1 ( select * from t1);
commit;
insert into t1 ( select * from t1);
commit;
insert into t1 ( select * from t1);
commit;
create index oid_ix on t1 (object_id) tablespace users ;
exec dbms_stats.gather_table_stats(null,'t1',cascade=>true, estimate_percent=>100);
sqlsql--
In my case the Table and Index looks this way:
JDBC@toekb> select table_name, NUM_ROWS,BLOCKS,AVG_SPACE from user_tables;
TABLE_NAME NUM_ROWS BLOCKS AVG_SPACE
=======================================
T1 485984 6944 0
Elapsed: 00:00:00.11
JDBC@toekb> select INDEX_NAME,BLEVEL,LEAF_BLOCKS,DISTINCT_KEYS,NUM_ROWS from user_indexes;
INDEX_NAME BLEVEL LEAF_BLOCKS DISTINCT_KEYS NUM_ROWS
===================================================
OID_IX 2 1074 60745 485960
Elapsed: 00:00:00.07
The table holds 7 times more blocks than the index !
any reply welcome
best regards
Edited by: guenterp on Aug 12, 2010 2:44 PM

Answer 1

guenterp wrote:
Hi,
does someone have a explanation for the following scenario:
I have a table T1 with an index OID_IX on column ( object_id ) - The table is a CTAS from dba_objects just to populate it with data.
There are no other Indexes present. The table and the index are analyzed !
When I run the following query the table is accessed FULL ( not using the index )
SELECT OBJECT_ID FROM T1;
SQL> select object_id from t1;
485984 rows selected.
Elapsed: 00:00:01.76
Execution Plan
Plan hash value: 3617692013
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 485K| 2372K| 1528 (1)| 00:00:19 |
| 1 | TABLE ACCESS FULL| T1 | 485K| 2372K| 1528 (1)| 00:00:19 |
Statistics
1 recursive calls
0 db block gets
7396 consistent gets
0 physical reads
0 redo size
2887158 bytes sent via SQL*Net to client
5684 bytes received via SQL*Net from client
487 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
485984 rows processed
But if I add a predicate ( even though it is useless in this case ) the index is taken and the query runs faster:
JDBC@toekb> select object_id from t1 where object_id != -999;
485960 rows selected.
Elapsed: 00:00:01.40
Execution Plan
Plan hash value: 3555700789
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 485K| 2372K| 242 (3)| 00:00:03 |
|* 1 | INDEX FAST FULL SCAN| OID_IX | 485K| 2372K| 242 (3)| 00:00:03 |
Predicate Information (identified by operation id):
1 - filter("OBJECT_ID"<>(-999))
Statistics
1 recursive calls
0 db block gets
1571 consistent gets
0 physical reads
0 redo size
2766124 bytes sent via SQL*Net to client
5684 bytes received via SQL*Net from client
487 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
485960 rows processed
here my setup:
sqlsql--
drop table t1 purge ;
create table t1 tablespace users as select * from dba_objects;
insert into t1 ( select * from t1);
commit;
insert into t1 ( select * from t1);
commit;
insert into t1 ( select * from t1);
commit;
create index oid_ix on t1 (object_id) tablespace users ;
exec dbms_stats.gather_table_stats(null,'t1',cascade=>true, estimate_percent=>100);
sqlsql--
In my case the Table and Index looks this way:
JDBC@toekb> select table_name, NUM_ROWS,BLOCKS,AVG_SPACE from user_tables;
TABLE_NAME NUM_ROWS BLOCKS AVG_SPACE
=======================================
T1 485984 6944 0
Elapsed: 00:00:00.11
JDBC@toekb> select INDEX_NAME,BLEVEL,LEAF_BLOCKS,DISTINCT_KEYS,NUM_ROWS from user_indexes;
INDEX_NAME BLEVEL LEAF_BLOCKS DISTINCT_KEYS NUM_ROWS
===================================================
OID_IX 2 1074 60745 485960
Elapsed: 00:00:00.07
The table holds 7 times more blocks than the index !
any reply welcome
best regards
Edited by: guenterp on Aug 12, 2010 2:44 PMSorry, but I am in doubt with your statists:
Statistics
1 recursive calls
0 db block gets
7396 consistent gets
0 physical reads ---->>>>>>>>>>>> Why it is 0 in any case(bnoth full scan and index FFS)
0 redo size
2887158 bytes sent via SQL*Net to client
5684 bytes received via SQL*Net from client
487 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
485984 rows processedCould you pls retry below operation before execuion of each sql:
alter system flush buffer_cache;In my case with 10.2.0.4 on HPUX:
SQL> exec dbms_stats.gather_table_stats(null,'t1',cascade=>true, estimate_percent=>100);
PL/SQL procedure successfully completed.
Elapsed: 00:00:03.61
SQL> alter system flush buffer_cache;
System altered.
Elapsed: 00:00:14.00
SQL> select * from t1;
82016 rows selected.
Elapsed: 00:00:06.48
Execution Plan
Plan hash value: 838529891
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT  |      | 82016 |  6888K|   394   (1)| 00:00:05 |
|   1 |  TABLE ACCESS FULL| T1   | 82016 |  6888K|   394   (1)| 00:00:05 |
Statistics
          0  recursive calls
          0  db block gets
       6455  consistent gets
       1039  physical reads
          0  redo size
    3480570  bytes sent via SQL*Net to client
      38508  bytes received via SQL*Net from client
       5469  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
      82016  rows processed
SQL> alter system flush buffer_cache;
System altered.
Elapsed: 00:00:18.26
SQL> select * from t1 where object_id !=999;
81976 rows selected.
Elapsed: 00:00:07.09
Execution Plan
Plan hash value: 838529891
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT  |      | 81976 |  6884K|   394   (1)| 00:00:05 |
|*  1 |  TABLE ACCESS FULL| T1   | 81976 |  6884K|   394   (1)| 00:00:05 |
Predicate Information (identified by operation id):
   1 - filter("OBJECT_ID"<>999)
Statistics
          0  recursive calls
          0  db block gets
       6443  consistent gets
       1039  physical reads
          0  redo size
    3478961  bytes sent via SQL*Net to client
      38494  bytes received via SQL*Net from client
       5467  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
      81976  rows processed
SQL>